Research On Geomagnetic Grid Positioning Model

Location information is very important to people whether in daily life or production activities.Currently,the outdoor positioning technology is mature,but the accuracy of the GPS positioning system in the indoor couldn't meet people's demand.In this paper,we propose a positioning system by the geomagnetic field signal guarantees both accuracy of positioning and cost reduction.This paper analyzes the basic characteristics of geomagnetic field,and discusses the basic model and positioning algorithm for positioning using geomagnetic information.When in the offline phase,the geomagnetic data in the area is collected,and these data is fitted with an approximate Gaussian distribution.Then the Gaussian filtered geomagnetic data is stored in the fingerprint database.When in the online phase,the collected data is calculated with the data in the database to obtain the corresponding confidence level.In this paper,several classic positioning fingerprinting algorithms are studied and experimented.The nearest neighbor method,K-nearest neighbor method and K-weighted nearest neighbor method are used.Then,a path matching method is proposed.Combining with the pedestrian track speculation,the motion trajectory is obtained from the heading estimation,the step number detection and the step size estimation.Finally,the concept of probability distribution was introduced,and an algorithm based on naive Bayes classifier was used for positioning analysis.The positioning algorithms proposed in this paper can obtain different precision of the results.Through experimental verification of these algorithms,it can be basically determined that the positioning algorithm based on naive Bayes classifier is better than the other methods in terms of accuracy.